INDEX

Explanations

words related to various academic and specific institutions

oai_token-act-pair · gpt-3.5-turbo

terms related to complex analysis and academic terminology

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 antioxid

-1.01

Balt

-0.86

RESULTS

-0.72

 behavi

-0.72

 corrid

-0.66

âĢ¢âĢ¢âĢ¢âĢ¢

-0.66

 ingred

-0.65

PDATE

-0.65

IDA

-0.65

 adolesc

-0.65

POSITIVE LOGITS

ividual

0.87

ogun

0.77

itially

0.77

 ItemLevel

0.75

enfranch

0.70

othal

0.67

ecause

0.66

rue

0.65

 BaseType

0.64

ageddon

0.63

Activations Density 2.297%

words related to various academic and specific institutions

terms related to complex analysis and academic terminology

No Known Activations